Pulse properties unit test fails

[Seabear-attack]

test_light.py fails at line 96 when calculating the time-bandwidth product. This fails for multiple pulse types, including gaussian and sech2 by about 10%. This has persisted since before the FFT backend change, and does not seem to change whether using FFTW3 or scipy.

[Seabear-attack]

Gaussian tbp: .480 (should be .441)
Sech2 tbp: .320 (should be .315)

[Seabear-attack]

Calculation of spectrum FWHM in light.py includes an extra added dv in line 510. Same thing in line 752 when calculating time-domain FWHM. Would explain the over-estimation of the TBP.

[Seabear-attack]

Calculation of spectrum FWHM in light.py includes an extra added dv in line 510. Same thing in line 752 when calculating time-domain FWHM. Would explain the over-estimation of the TBP.

Removing this term changes the TBP estimates such that they're much more wrong. I've determined this to be somewhat of a sampling issue, as cranking up the number of points improves agreement with the theoretical TBP dramatically. I'm concluding this as a non-issue and updating the test to have higher tolerance.